Академический Документы
Профессиональный Документы
Культура Документы
Low pressure or atmospheric storage tanks are commonly used for the storage of
flammable hydrocarbons. In order to minimise losses of volatile organic compounds
to the environment, these tanks are often fitted with pressure / vacuum valves
(PV valves). These enable the tank to operate at pressures typically between +56mbar
and -6mbar before the valve opens.
The atmosphere within the tank is often in the flammable region and hence during
outbreathing, the vapours vented from the PV valve are also flammable. These may be
ignited by an external ignition source such as static discharge or hot work occurring
near the vent. The common assumption has always been that the tank would be
protected from internal explosion by the presence of the PV valve on the basis that the
flame could not pass back through the valve.1
This paper presents the opinion that ignition can, in fact, flash back through a PV
valve based on consideration of flame speed, minimum experimental safe gap (MESG)
and flow rates. Experimental work done by Protego will also be presented demonstrat-
ing that flashback can occur, especially under conditions of continuous venting. It is
therefore recommended that consideration should be given to installing a flame arrester
on tanks protected by a PV valve.
INTRODUCTION
Low pressure or atmospheric storage tanks are often used for the storage of flammable
materials i.e. those with flashpoints of less than about 50C. Thus, under normal operating
conditions it is normal for there to be a flammable atmosphere inside the tank. In order to
minimise vapour losses from the tank,
Taking into consideration the normal operation of LP storage tanks, it can clearly be
seen that vapour will be periodically vented from the tank to atmosphere. The volume and
flowrate of the emission will be dependent upon a number of factors such as:
l Pumping in (filling) rate
l Diurnal temperature changes
l Internal heating
Symposium Series NO. 154 2008 IChemE
Under certain conditions, flammable vapour may be vented from the tank for consid-
erable periods, especially for larger tank sizes. Thus, there will be a cloud of flammable
material vented, generally at high level, over the roof space of the tank.
In particular, during filling of a tank with a flammable liquid e.g. Hexane or
similar, it is possible that a flammable mixture could be vented for a period of more than
30 minutes.
From
tank
Fill funnel
Liquid seal
Drain
FLASHBACK PROTECTION
It has always been assumed, at least within the UK, that protection against the flashback of an
ignition into the tank is provided by the provision of a Pressure / Vacuum Valve (PVV) on the
vent line. This has generally been endorsed by the HSE5 with the result that the vast majority
of tanks containing flammable materials which are equipped with PV valves do not have asso-
ciated flame arresters in the vent line. A typical PV Valve configuration is shown in Diagram 2
below. Operation of the valve under pressure and vacuum are shown in diagrams 3 & 4.
Theoretically, in order for the protection to be effective, one of two conditions must
be met
as follows:
Either:
The velocity through the valve must be greater than the turbulent flame velocity
Or:
The gap between the valve pallet and the seat must be less than the Maximum
Experimental Safe Gap (MESG)
Some typical MESG values are:
Propane 0.965
Methane 1.14
Ethylene 0.65
Hydrogen 0.50
Hexane 0.95
Cyclohexane 0.94
Ethyl acetate 1.04
Symposium Series NO. 154 2008 IChemE
It can be seen that these values are a maximum of just over 1mm.
Note: The MESG is the largest gap through which a flame will not pass. This param-
eter is used in the specification of flame arresters to ensure that a flame will not pass
through the arrester. MESG is measured for stoichiometric mixtures (the worst case).
Taking a typical 3 (75mm) diameter PV valve, the opening for the pallet is in the
order of 2 (50mm) at a flow in the order of 600 m/h6. It can be seen that the opening is
far greater than the MSEG and thus it must be assumed that, if the flame has sufficient
velocity, it may pass back through the PV Valve.
Symposium Series NO. 154 2008 IChemE
A 2 (50mm) pallet lift gives a superficial gas velocity in the order of 14 m/s through
the valve based of Protego flow capacity charts for the DN80/3 valve6, although this could
be significantly less for lower tank design pressures where higher lift is required for to
achieve the desired flow under normal operating conditions. It can be seen that this is a
relatively low velocity for turbulent flame speed. In fact, the superficial gas velocity
through the valve could be as low as 2-3 m/s with a lift of 5mm or more, considerably more
than the MESG.
the gas flow is as laminar as possible, including a very smooth tube wall and a long,
straight upstream tube.
Laminar flame speed, is, however, of little use in a real-world situation such as
lowpressure tank vents because the nature of the flows and equipment that are in use are
far more likely to result in turbulence within the gas mixture. The key factor in the trans-
mission of the flame is the turbulent flame speed. The turbulence of the gas mixture has a
significant effect on the flame speed as does the degree of confinement. Turbulent mixtures
in confined spaces (such as pipelines) can reach speeds of 300 m/s.
Turbulence in the gas will increase the flame speed since the size of the flame front
will be increased by eddies resulting in increased reaction rate. Increased reaction rate and
flame speed form a positive feedback loop by causing an increase in turbulence and hence
the flame can accelerate. In extreme cases, flame speeds of greater than 1000 m/s can be
reached in long pipelines where deflagration can transition to detonation.
In practice, under real-world conditions, there is always a degree of turbulence asso-
ciated with venting from tanks. Considering a typical LP tank vent system configuration it
is clear that under many of the flow conditions that may commonly occur the flow through
the system will be turbulent. The turbulence will be caused partially by the changes in
section and direction within the system and partly related to the velocity (as the Reynolds
number increases).
Note that these maximum velocities are achieved by near-stoichiometric mixtures
and flame velocities do fall off away from stoichiometric mixture. However, flame veloci-
ties are still high enough to be greater than the flow velocity through the vent system
(including the PV valve).
PROTEGO TESTS
Tests done on conventional PVVs by Protego in Germany indicate that PVVs may, in
some circumstances allow an ignition to pass through the valve into the tank. Obviously,
this would result in an internal explosion and destruction of the tank. Three tests are shown
with this paper, these are as follows:
that the pressure outside the valve caused by the burning gas can lift the vacuum pallet
against its weight thus allowing passage of the flame, pressure outside the valve being
equivalent to a vacuum inside.
Flammable mixture
Ignition
location
Plastic bag
Membrane
Flammable mixture
It is known that vapour cloud explosion do not produce a great deal of pressure unless
in congested or confined regions, it is, however, foreseeable that an energetic (near stoichio-
metric) gas mixture might foreseeably generate more than 10 mbar overpressure in the
region of the valve, especially if the region around the valve is congested by pipework etc.
The issue here is that ignition of a reasonably sized vapour cloud outside of and surrounding
a PV Valve may foreseeably result in the generation of sufficient pressure to lift the vacuum
pallet. This will then provide a clear route for the flame back into the vessel.
Endurance Burning
If vented material from the PV Valve continues to burn for an extended period then it is
foreseeable that the valve cap, pallet etc will become hot. If this continues to the point at
which the temperature of the metal reaches the autoignition temperature of the vapour then
a flashback may occur. This test shows flashback through the valve occurring after approx-
imately six minutes. Note that during this test the flame was not played directly onto the
valve but was directed to one side and had no significant effect on the heating of the valve.
Symposium Series NO. 154 2008 IChemE
Thus, heating of the valve occurred solely as a result of the burning of the gas coming out
of the PV Valve.
The reason for the ignition of the gas so quickly is that the pressure pallet is of light
construction, as required for the duty, and thus will heat up comparatively rapidly. In the
video it can clearly be seen that the steel valve bonnet reaches red heat very quickly. This
item is generally constructed of light gauge steel since it is only there to provide weather
protection. This kind of temperature will also rapidly destroy any polymeric seals on the
pallet destroying the valve integrity.
It is noted that this circumstance does rely on the sustained ignition of the gas
mixture leaving the valve and the resultant heating of the valve material. It is, however,
foreseeable that there will be circumstances, such as during tank filling, where venting
may occur for an extended period.
Thus, this is another event that may be considered credible, if a rare event.
Flame
Since the pallet lift is clearly greater than the MESG of the mixture, the mechanism
for transmission must be the turbulent burning velocity of the gas being greater than the
superficial gas velocity leaving the valve.
This is considered to be a credible event since the combination of pallet lift and an
external ignition source are a foreseeable combination.
Flame
Vented
gas
CONCLUSIONS
It can be seen that the installation of a PV Valve on a low-pressure storage tank containing
a flammable vapour does not provide total protection against the flashback of an external
ignition into the tank. The combination of the lift of the PV Valve pallet (which will, in the
vast majority of cases be greater than the MESG of the vapour), the turbulent flame speed
and the superficial gas velocity coming out of the valve can result in circumstances where
it is theoretically possible for the flame to pass into the vessel.
Based on the tests conducted by Protego, is is clear that this theoretical considera-
tion is, in fact, borne out in reality under test conditions. The test conditions used, whilst
they might be considered to be arduous, are, in fact credible, especially in the third
case where the ignition of the gases venting from the valve results in flashback into
the vessel.
Equally of concern is the flashback through the valve occurring when the valve is
closed (the first case) where ignition of an external flammable cloud flashes back through
the vacuum pallet.
Thus, it is concluded that there is a small but significant risk of flashback through a
PV Valve under conditions of external ignition. On this basis, consideration should be
given to the provision of a flame arrester in the vent system in order to provide additional
Symposium Series NO. 154 2008 IChemE
Flame arrester elements shown in blue. Note that other configurations are possible
Diagram 5. PV valve with integral flame arresters
flashback protection. The flame arrester could be located either at the inlet to the valve
(between the valve and the tank, Diagram 5) or else on the outlet of the valve.
REFERENCES
1. BS EN 12874: 2001 Flame arresters
2. API 650 Welded Steel Tanks for Oil Storage, 10th Edition
3. EN 14015:2004 Specification for the design and manufacture of site built, vertical,
cylindrical, flat-bottomed, above ground, welded, steel tanks for the storage of liquids
at ambient temperature and above
4. API 620 Design and Construction of Large, Welded, Low-Pressure Storage Tanks,
10th Edition
5. HSE Contract Research Report CRR281 Investigations into concerns about BS EN
12874:2001 flame arresters; HSE 2004
6. Protego Pressure/Vacuum Relief Valve Catalogue for type VD/SV-HR
7. Garstein M, Levine O & Wong EL; Fundamental flame velocities of hydrocarbons;
Ind Eng Chem V43, pp2770-2772; 1951
8. Coward HF, Jones GW; Limits of flammability of gases and vapors; USBM Report
No.503, 1952
10